Advances in the Design of the icub Humanoid Robot: Force Control and Tactile Sensing Bled, Slovenia October 26 th, 2011 Giorgio Metta, Ugo Pattacini, Andrea Del Prete, Marco Randazzo, Alberto Parmiggiani, Serena Ivaldi, Matteo Fumagalli, Marco Maggiali, Lorenzo Natale, Francesco Nori, Giulio Sandini Cognitive Humanoids Laboratory Robotics, Brain and Cognitive Sciences dept. the Italian Institute of Technology
what are we looking for? the focus of our research is in the implementation of biologically sound models of cognition in robots of humanoid shape this has the two-fold aim of: furthering our understanding of brain functions realizing robot controllers that can learn and adapt from their mistakes
we set up to reach our goals by designing a humanoid robot platform, namely the icub making it the platform of choice for researchers in artificial cognitive systems studying cognition from a developmental perspective (neuroscience)
icub is an open source international endeavour initially funded by the EU project RobotCub a full humanoid robot is 104cm, weighs 22 kg has 53 degrees of freedom can crawl, sit and manipulate open design as LGPL/GPL
why is the icub so special (for us)? hands: we started the design from the hands 5 fingers, 9 degrees of freedom, 19 joints sensors: human-like, e.g. no lasers cameras, microphones, gyros, encoders, force, tactile electronics: flexibility for research custom electronics, small, programmable (DSP) reproducible platform: community designed reproducible & maintainable yet evolvable platform
force/torque control on the icub joint torque sensors pros: direct feedback loop cons: requires mechanical re-design six-axis F/T sensors pros: scalability, full perception cons: computational delays
icub sensorization
0
Test of the dynamical model
icub 2.0 new mechanics compliant actuators force/torque measurements
icub sensorization new icub arm: integrated joint torque sensor on majors joints: shouder (3 DoF) + elbow (1 DoF) Shoulder roll Shoulder pitch Elbow semiconductor strain gauges (SSGs)
icub Hardware: elbow joint Elbow (1DOF): 2 SSGs configured as an half Wheatstone bridge.. Calibration: elbow c1 ( s1 o1)
icub Hardware: shoulder roll Shoulder (3 DOF): Shoulder Roll: 4 SSGs configured as two half Wheatstone bridges.. Calibration: roll c2 ( s2 o2) c3 ( s3 o3)
icub Hardware: shoulder pitch Shoulder (3 DOF): Shoulder Pitch: 4 SSGs mounted directly on an hollow motor shaft..coupling with the shoulder roll: 1 1 0 T T m T j T 0 1/ r 1/ r 0 0 1/ r Calibration: pitch c ( s o ) c ( s o ) 4 4 4 5 5 5 roll
icub Hardware: shoulder yaw Shoulder (3 DOF): Shoulder Yaw: No additional joint torque sensors required. The joint torque is obtained through the direct measurements of the six axis F/T sensor. X Z Y. No calibration is required (the F/T sensor is already calibrated): yaw F y d T z
joint torque sensors
experiments and model validation static configuration: an additional six axis F/T sensor is placed at the end effector to measures the external wrenches w e in this experiment we consider the following quantities: joint torques measured by the joint torque sensors: t j joint torques computed from the arm F/T sensor: t FT joint torques estimated thought the additional F/T sensor located at the end effector: t e =J T w e joint torques predicted by the arm model (no external forces): t m additional F/T sensor at the end-effector Joint 0 Joint 1 Joint 2 Joint 3 E(τ j -τ ft ) 0.127 Nm -0.049 Nm -0.002 Nm -0.032 Nm σ(τ j -τ ft ) 0.186 Nm 0.131 Nm 0.013 Nm 0.042 Nm E(τ j -(τ m +τ e )) 0.075 Nm -0.098 Nm -0.006 Nm 0.006 Nm σ(τ j -(τ m +τ e )) 0.191 Nm 0.173 Nm 0.020 Nm 0.032 Nm
experiments and model validation representation of the external wrenches: the arm FT sensor allows to retrieve also the external wrench at the end effector. w e : the external wrenches measured by the additional F/T sensor at the end-effector w FT : the external wrenches computed using the arm F/T sensor remarks: it is not possible to estimate the externally applied wrenches w e ϵ R 6 using the only measurements of joint torques t j ϵ R 4 joint torques are effected by the null space of the Jacobian F 1 (N) F 2 (N) F 3 (N) μ 1 (Nm) μ 2 (Nm) μ 3 (Nm) E(w e -w FT ) 0.181-0.465-0.154-0.079-0.024-0.024 σ(w e -w FT ) 0.384 0.426 0.469 0.149 0.048 0.059
force/torque control on the icub icub 2.0
skin principle lots of sensing points structure of the skin
fingertips capacitive pressure sensor with 12 sensitive zones 14.5 mm long and 13 mm wide, sized for icub embedded electronics: twelve 16 bit measurements of capacitance either all 12 taxels independently at 50 Hz or an average of the 12 taxels at about 500 Hz
touch
some philosophy to conclude why open (source) platforms? repeatable experiments benchmarking quality this also resonates with industry-grade R&D in robotics
sponsors EU Commission projects: RobotCub, grant FP6-004370, http://www.robotcub.org CHRIS, grant FP7-215805, http://www.chrisfp7.eu ITALK, grant FP7-214668, http://italkproject.org Poeticon, grant FP7-215843 http://www.poeticon.eu Robotdoc, grant FP7-ITN-235065 http://www.robotdoc.org Roboskin, grant FP7-231500 http://www.roboskin.eu Xperience, grant FP7-270273 http://www.xperience.org EFAA, grant FP7-270490 http://efaa.upf.edu/ More information: http://www.icub.org